Research On Statistical Soil Properties And Reliability Of Soil Slope Stability

Considerable statistical data is required in reliability analysis. In 1980s, the calculated reliability index was too low since the variability of soil properties was incorrectly estimated. A set of practical research methods has been formed from statistical soil properties, and separate factor determination to reliability index modification by more than 20 years, and it has gone through stages of foundation and development research. As to statistical soil properties, particularly great proceeding has been achieved in application of stochastic field theory, including statistical correlation distance and shear strength index, and calculation of spatial mean value and variation coefficient. But there exists unbalance in different districts, and it still has a long way to go to practical engineering. Although the standard of Shanghai soil and foundation design has firstly input reliability method, application on slope stability evaluation has not been put in standard yet. The thesis has made a systematic study and summarize from choosing characteristic parameters needed by reliability analysis to reliability analysis method of soil slope.Uncertainty sources of soil properties were analyzed. Since there is spatial correlation in soil properties, stochastic field theory was explored in studying spatial mean value and spatial variance combining with soil probabilistic model. As to data of test or monitoring, its reliability is to be verified. In the thesis a few of prediction techniques to deal with uncertainties were summarized, which has significance to statistical little sample or poor test data. When reliability analysis being performed, optimization approach based on simplified Bishop method and Chen-Morgenstera extension general slices method were chosen and accomplished in Microsoft Excel, which has provided a simple, intuitive and practical way for reliability analysis. Two search algorithms to achieve probabilistic critical slip surface are based on reliability index. In addition, three trains of thought and development tendency of soil slope reliability analysis are also proposed.The studying approaches and main results can be summarized as follows:1. Uncertainty sources of soil properties are analyzed, and the formulations of spatial mean value and variation coefficient are given out. Statistics method and noticeable items are explained in the example.2. Various computation methods of correlation distance are compared with each other. The effects of different soil stratum and various soil properties on correlation distance are studied. It comes out to be that the correlation distance is a kind of soil property, and different soil stratum has its own correlation distance. The correlation distance derived from various soil parameters in the same stratum has no distinct difference, although the value resulted from in-situ tests which are considered to be more proper, is generally greater than that from laboratory tests. In the thesis the mean value and range of correlation distance resulted from in-situ tests are listed for reference. It could be helpful in the arrangement of sample interval, sample location and evaluation of spatial variance.3. Types of errors in test or monitoring data are analyzed and verifying methods are suggested. For example, Grubbs criterion and t-test can be used to deal with mistaken error, and they are suitable to little sample. Statistics test methods that are suitable to geotechnical engineering are also suggested in the thesis. For example, t-test can be used to identify if two group of data come from the same origin. If they are, they can be put together for statistics, orthey should be made respectively. Variance analysis was used in chapter 3 to study the effects of different soil stratum and soil properties on the correlation distance, and some valuable conclusions were achieved.4. Uncertainties can be divided into four kinds: stochastic, fuzzy, grey and unknown, and probabilistic statistical method, fuzzy mathematics, grey system and evidence theory have been developed to dea...